Minemizer Benchmark Report

Token Efficiency Summary

Normalized (JSON pretty = 1.0x, higher is better)

Tokenizer × Format

Average efficiency per tokenizer (normalized)

LLM Accuracy Summary

Nested Data

7 dataset(s), 6 model(s). Sorted by efficiency.

Flat Data

2 dataset(s), 6 model(s). Sorted by efficiency.

Sparse Data

1 dataset(s), 6 model(s). Sorted by efficiency.

Efficiency = Accuracy × (JSON tokens ÷ Format tokens). Higher is better — balances accuracy with token savings.